Current Research (2003) Newfoundland Department of Mines and Energy

Geological Survey, Report 03-1, pages 223-229

GUIDELINES FOR THE CALCULATION AND USE OF SULPHIDE METAL

CONTENTS IN RESEARCH AND MINERAL EXPLORATION
A. Kerr
Mineral Deposits Section

ABSTRACT

The metal contents of samples of magmatic sulphide mineralization are generally correlated with their sulphide contents,
because the metal content of bulk sulphides remains approximately constant in this environment. Sulphide metal contents pro -
vide valuable information in mineral exploration and research, and are readily calculated if the mass fraction of sulphides is
known. The latter is most simply calculated from the measured sulphur content using an assumed sulphur content of 35.7% S
for a typical pyrrhotite–chalcopyrite–pentlandite mixture. There appears to be little advantage in accounting for the individ-
ual sulphide minerals in the sample. For samples containing > 10% S, calculations are very robust, but lower sulphur con-
tents generally demand at least some correction for metals contained in nonsulphide minerals. These can become significant
below 5% S, and/or in olivine-rich samples. Corrections can readily be performed by simple mass-balance calculations, using
concentration data from unmineralized host rocks.

Significant uncertainties are introduced by analytical errors for sulphur and metals, which are commonly measured from
separate sample aliquots. The combined errors in calculated sulphide metal contents will generally exceed ± 10%, but get
larger at low S contents. Caution must generally be used in the treatment of data from samples containing < 2.5% S, espe-
cially for the PGE. Application of the method in simple grade-potential assessment is easy, but research-oriented studies
involving sulphide-poor samples are more problematic. Invalid correction of data for nonsulphide-hosted metals can lead to
false negative or positive correlations between metal contents and sulphide content, which itself could be linked to geologi-
cal parameters, such as depth within an intrusive body. There are also several valid geological reasons for such correlations,
but data of this type require careful assessment to separate true variation from artifacts. Propagated analytical uncertainties
are largest in sulphide-poor samples, for which corrections may also be required. These potential inaccuracies must be borne
in mind whenever data from different localities or mineralization types are compared and contrasted. Simple charts allow
assessment of the reliability of data in various situations, and spreadsheet programs were developed for easy recalculation of
data.

INTRODUCTION phides, and thus provide an upper limit for undiscovered

In magmatic sulphide deposits, such as Voisey’s Bay,
ore-metal concentrations are typically strongly correlated
with the sulphide content of samples. In many producing
deposits, such as those of the Sudbury area, face samples are
rarely assayed, because visual estimates of the amounts of
sulphide content suffice for grade-control purposes. This
correlation reflects the fact that essentially all base metals
and platinum-group-elements (PGE) are contained in sul-
phide minerals, and that the bulk sulphide fraction shows a
narrow range of values for these elements, within a given
deposit. This parameter is called the sulphide metal content
or tenor; in this article the first term is preferred, because
“tenor” can also refer to the proportion of sulphides within
an ore sample. Calculated sulphide metal contents represent
a grade projection for material consisting of 100 percent sul-
massive sulphide deposits on a given property. However,
they provide no inference as to the actual existence of such
deposits! Sulphide metal contents also permit the examina-
tion of subtler variations in metal contents and metal ratios
from a more academic perspective (e.g., Lightfoot et al.,
2001a, b). Perhaps the most useful application of calculated
sulphide metal contents is in comparing and classifying
magmatic sulphide mineralization, by removing “noise”
linked to variations in sulphide content. The procedure is
quite common in exploration, and increasingly popular in
research applications, although there are few formal descrip-
tions and evaluations of it. This article outlines the proce-
dure and provides some guidelines for the use of such cal-
culations in different types of studies. For a more detailed
discussion, readers are referred to a forthcoming article in
Exploration and Mining Geology (Kerr, in press).

223
CURRENT RESEARCH, REPORT 03-1
SUMMARY OF CALCULATION METHODS
BASIC PRINCIPLES
In a typical sample of magmatic sulphide mineraliza-
tion, virtually all of the base metals and PGE reside in the
sulphides. Nickel and Co are generally in pentlandite, and
Cu in chalcopyrite, but all the metals were originally dis-
persed in a high-temperature sulphide mineral termed
Monosulphide Solid Solution (MSS), which converted to
pyrrhotite, pentlandite and chalcopyrite as it cooled. PGE
may be hosted in chalcopyrite or pentlandite, but they may
also occur in a wide variety of rarer sulphide, telluride and
arsenide minerals. The main sulphide minerals have similar
sulphur contents, i.e., 33.3% (pentlandite), 34.9% (chal-
copyrite) and 36.5% (pure FeS - i.e., troilite; natural
pyrrhotite will be more S-rich, as it is metal-deficient). If all
metals are in sulphide minerals, the sulphide metal contents
can readily be derived from the whole-rock analysis, pro-
viding that the mass fraction of sulphides is known. The lat-
ter can be derived by dividing the sulphur content of the
sample by the assumed sulphur content of a rock consisting
purely of sulphides. As the dominant sulphide mineral is
usually pyrrhotite, a common approach is to assume a value
of 36.5% S for pure sulphides. The sulphide metal content is
then obtained by dividing the whole-rock metal content by
the mass fraction of sulphides. For example, if a sample con-
tains 4000 ppm Ni, and 7.3% sulphur, the mass fraction of
sulphides is 0.2 (i.e., 7.3 / 36.5), and the sulphide metal con-
tent is 20 000 ppm (i.e., 4000 / 0.2), or 2% Ni.
All calculations involve this basic method, which is
simplistic, because some metals can reside in silicates, and
the sulphides include more than one mineral. The procedure
is also commonly an extrapolation because the samples may
actually contain only small amounts of sulphides. Extrapo-
lations are prone to error propagation, and there are several
potential sources of errors, which also need to be considered
in interpreting data.
CALCULATING THE MASS FRACTION OF
SULPHIDES
Methods for calculating the mass fraction of sulphides
differ only in the assignment of an assumed sulphur content
for pure sulphides. There are essentially two approaches,
i.e., to use an arbitrary value (as outlined above), or to
endeavour to account for the sulphide minerals individually.
In previous articles (e.g., Kerr, 1999), the author
assigned a value of 35% S for pure sulphides. This is an
underestimate, as pyrrhotite (>36.5% S) is generally the
dominant sulphide. This choice was deliberately made to
ensure that estimates of sulphide metal contents remain con-
servative. Naldrett et al. (2000) and Li et al. (2001) summa-
224
rize the more complex methods. In these, Ni and Co are
combined, and assigned to pentlandite. The latter is assumed
to have a combined (Ni+Co) content of 36%, typical of
material from Sudbury. The sulphur required to make pent-
landite is then subtracted from the total sulphur content.
Copper is then assigned to chalcopyrite, and the total sul-
phur content is again corrected. The remaining sulphur is
then assigned to pyrrhotite, using either the ideal FeS for-
mula (Naldrett et al., 2000) or Fe7S 8 (Li et al., 2001). The
sulphur content of a rock containing 100 percent sulphides
is then readily calculated from the proportions of the three
sulphide minerals and their ideal sulphur contents. This
method also results in a slight underestimate if “ideal” FeS
is used, but it may result in an overestimate, if Fe7S8 is used.
Calculations suggest that differences between these
methods are trivial. Typically, the differences in calculated
sulphide metal contents are less than 0.1 wt%, and are insen-
sitive to Ni/Cu variations, except at very low sulphide con-
tents. A close fit to the method of Naldrett et al. (2000) was
obtained by using a slightly revised value of 35.7% for
100% sulphides. Compared to the uncertainties introduced
by analytical errors, and those involved in correction for
nonsulphide-hosted metals, such differences are negligible.
There thus appears to be no advantage in the more complex
approach.
CORRECTIONS FOR METALS IN NONSULPHIDE
MINERALS
There are instances where essentially all metals reside
in sulphide minerals, as for example in discordant sulphide
mineralization hosted by monomineralic anorthosites. How-
ever, in most cases, some metals will reside in silicate min-
erals such as olivine and pyroxene, and perhaps also in iron
oxides. In particular, olivine-bearing rocks are likely to con-
tain significant silicate-hosted Ni. Under some circum-
stances, corrections must be made for these metals.
Mass-balance equations dictate that :
X WR = (Xsulph * Msulph) + (X other * M other ) (1)
where X indicates the concentrations (in whole-rock, sul-
phides, and other minerals, respectively) and M indicates
the relevant mass-fractions of each. Rearrangement of equa-
tion (1) shows that the true sulphide metal content is given
by :
X sulph = [ XWR - (X other * M other ) ] / M sulph (2)
Thus, corrected sulphide metal contents are readily obtained
if the concentration in bulk nonsulphides is known, because
M other is given simply by [1 - Msulph].
 A. KERR

Sulphide metal contents used in previous articles (e.g., Co Ni, Cu

Kerr, 1999; Kerr and Smith, 2000) were corrected via this 0
method, using values for Xother based upon analytical data S
-10 %
from unmineralized host rocks of analogous composition. If 30
such data are unavailable, an arbitrary value can be substi- -20
tuted.

Correction (%)
-30

S
20%
-40
Nonsulphide metal corrections suggested by Naldrett et

S
1%
al. (2000) use essentially the same method, but account only -50

S
for Ni in olivines. The calculation of the Ni content of

1 0%

S
-60

0.1%
olivine is based on the observation that Ni/Fe(olivine) =

5% S
(Ni/Fe(sulphide) / 25) in samples from Voisey’s Bay, in which -70
the olivines are believed to have reequilibrated with sul- -80
phide liquid (Li and Naldrett, 1999). Assuming an olivine
-90
composition of Fo65, the Ni content of coexisting olivines
(X olivine), is obtained from the uncorrected sulphide metal -100
1 10 100 1000 10000
content. The method also requires an estimated value for the Xsulph / Xother
modal proportion of olivine (M olivine) in the nonsulphide por-
tion of the sample, from which the bulk nonsulphide Ni con-
Figure 1. Curves expressing the magnitude of corrections
tent is derived for use in equation (2). Calculations show
for metals contained in nonsulphide minerals as a function
that the magnitudes of corrections using this method resem -
of the amount of sulphur, and the ratio between the uncor-
ble those from the simpler method, providing that the modal
rected sulphide metal content (X sulph) and the estimated bulk
proportion of olivine lies in the range of 20 to 40%. How-
nonsulphide metal content (Xother). Typical relative values
ever, the magnitude of corrections increases at higher sul-
for Xsulph/Xother for the Voisey’s Bay deposit in Labrador are
phide metal contents, whereas it remains essentially fixed in
also indicated. See text for further explanation.
the first method.

It is evident from equation (1) that the contribution of
sulphides to the whole-rock metal contents will be over-
whelming if Msulph is significant because the concentrations
in sulphides will generally be at least two orders of magni-
tude greater than in the bulk nonsulphides. In rocks that con-
tain large amounts of sulphide and in which sulphide metal
contents are high, the contribution of metal from sulphides
is so overwhelming that any correction is unnecessary (see
below). At low sulphide contents (or if sulphide metal con-
tents are low), the first method has several advantages, in
addition to being simpler to calculate. First, it requires no
assumptions about olivine contents, compositions or Fe/Ni
partitioning effects. Second, it is more readily adaptable to
local circumstances.
THE IMPORTANCE OF NONSULPHIDE METAL
CORRECTIONS
Nonsulphide metal corrections are inversely correlated
with the mass fraction of sulphides, and become most sig-
nificant as M sulph approaches 0. They are also inversely cor-
related with the ratio between the metal contents of sul-
phides and nonsulphides (X sulph/Xother ), and become less sig-
nificant at high sulphide metal contents.
These relationships are best illustrated by graphing the
required correction (here expressed as a negative percentage
of the uncorrected sulphide metal content) and the ratio
X sulph/Xother (where Xsulph is the uncorrected sulphide metal
content). Figure 1 shows several curves for variable
amounts of sulphur in a sample, ranging from 0.1% to 30%.
If the estimated value of X sulph/X other for a given sample does
not intersect the appropriate sulphur content curve, the cor-
rection exceeds !100 percent, and the procedure is mean-
ingless. This outcome may indicate that sulphides are metal-
free, but it more likely reflects overestimates of Xother . At low
S contents, corrections may be very significant unless the
sulphides are very metal-rich. Conversely, at high sulphide
contents, corrections are insignificant unless the sulphides
are very metal-poor. The graph in Figure 1 is useful in indi-
cating whether or not nonsulphide metal corrections are
needed in a given situation.
Figure 1 also shows that the interpretation of data from
sulphide-poor rocks that contain metal-bearing nonsulphide
phases could be difficult. In the absence of correction, there
will be an apparent inverse relationship between sulphide
metal contents and the amount of sulphide, even if the true
sulphide metal content is constant. Conversely, overly con-
servative corrections for sulphide-poor samples could lead
to false patterns in which corrected sulphide metal contents
are positively correlated with sulphur abundance. Sulphide
metal contents and sulphur abundance have been noted to be
inversely correlated (e.g., Barnes et al., 1997; Kerr, in

225
CURRENT RESEARCH, REPORT 03-1
press), but viable geological explanations exist for such pat-
terns. When dealing with sulphide-poor rocks, the sulphur
content of samples, and the precision of such data, must be
closely monitored. Also, reliable “background” data from
sulphide-free samples should routinely be acquired, so that
corrections are as accurate as possible.
The magnitude of required corrections will also vary for
different elements. In unmineralized mafic rocks, Ni and Co
commonly have similar abundances, but magmatic sulphide
deposits have much higher Ni/Co ratios, and the ratio
X sulph/Xother will generally be an order of magnitude greater
for Ni than for Co. Calculated sulphide Co contents from S-
deficient samples are thus unreliable, and other methods of
estimating this parameter should be used. Sulphide Cu con-
tents are the least sensitive to corrections, but this advantage
is balanced by more intrinsic “geological” variation, mostly
due to fractionation during crystallization of sulphide liq-
uids (e.g., Barnes et al., 1997).
ERRORS IN CALCULATED SULPHIDE
METAL CONTENTS
The results of any calculation are only as accurate or
reliable as the input data. In the case of sulphide metal con-
tent calculations, input errors come from analyses of metals,
and analyses of sulphur content. Furthermore, base metals,
PGE’s and sulphur are generally analysed using different
techniques, from discrete aliquots of the original sample.
Thus, sample heterogeneity may also be an important con-
sideration. As outlined above, nonsulphide metal corrections
become important at low sulphur contents, or when metal
concentrations are low, or when both situations apply. Thus,
all of these potential problems converge at low sulphur con-
tents and low concentrations.
Precision for sulphur analyses (Leco furnace method)
was investigated for within-batch duplicates at a single lab-
oratory and sample splits analyzed at two different laborato-
ries. Within-batch duplicates are precise to ± 4% or better at
all concentrations, but inter-laboratory precision is worse,
and at low S concentrations precision is as poor as ± 10 to
15%. Some samples containing > 20% S also exhibit poor
precision, suggesting that sample heterogeneity plays a role.
Given that sulphur analyses almost always involve an
aliquot separate from that used for metals, this suggests pre-
cision of ± 5% at all levels, degrading to ± 10% below about
2.5% S, and perhaps further degrading below 1% S.
Precision for base metals at the Department of Mines
and Energy (ICP-ES method) was assessed from almost 300
duplicate analyses undertaken between 1996 and 1998
(Finch, 1998; unpublished data). Average precision for all
data ranges from ± 5% for Ni to ± 6.5% for Cu. This
226
improves significantly at intermediate values typical of
weakly mineralized samples, ranging from ± 2.9% for Co (>
50 ppm) to ± 4.6% for Cu (> 100 ppm), but degrades again
above 1000 ppm. Thus, conservative estimates of precision
for base metals are around ± 5% to ± 6%.
Kerr (2002) recently presented results from a survey of
the PGE contents of Labrador magmatic sulphides, using
fire assay/ICP-ES analysis. Data from duplicate analyses
(separate sample splits) indicates that analytical precision is
poor below 100 ppb, likely worse than ± 10%, with very sig-
nificant degradation approaching detection limits. Above
100 ppb, an appropriate conservative estimate of precision is
at least ± 10%. Reanalysis of anomalous PGE values for
verification purposes at a different laboratory resulted in dif-
ferences in PGE contents of up to 50%, suggesting that sam-
ple heterogeneity and/or “nugget effects” from discrete Pt-
and Pd-rich mineral phases are also a potential problem
(Kerr, 2002).
The combined effects of analytical uncertainties for sul-
phur and metals upon calculated sulphide metal contents are
illustrated in Figure 2. This relates precision in metal analy-
sis to total combined uncertainties, with a series of lines
indicating variation in the precision of sulphur analysis. This
assumes that errors in metal contents and sulphur contents
are inversely correlated, which is unlikely, so these lines are
conservative estimates, but they are large. For example, pre-
cision of ± 10% for both sets of input data yield combined
uncertainties of +22.2% and -18.2% (mean 20.2%) in calcu-
lated sulphide metal contents. If the precision of sulphur
analysis improves to ± 2%, these drop to +12.2% and -
11.8% respectively (Figure 2a). The positive error is always
greater than the negative error because it involves greater
extrapolation, and the imbalance between the two grows
with increasing uncertainty. The absolute magnitudes of
errors can be quite high. For example, in a sample contain-
ing 1000 ppm Ni and 1% S, the total potential uncertainty in
sulphide Ni content is 1.4% Ni, which is clearly significant
in economic terms. Uncertainties in sulphide metal contents
that are corrected for nonsulphide metals are greater than for
uncorrected values, although this difference diminishes with
increasing sulphur and metal contents. There are of course
also uncertainties involved in the correction procedure itself,
but these are more difficult to quantify.
Figure 2b is useful as a general guide to the assessment
of potential total uncertainty as a function of analytical pre-
cision for metals and sulphur. Under most circumstances,
uncertainties of ± 12% or so should be expected. The poten-
tial uncertainties are greatest below 1% S, where precision
degrades, or if the analyses of metals are imprecise, as they
may be for PGE. Combined uncertainties of ± 40% are
entirely possible under these circumstances, and caution
 A. KERR

35% grades of potential massive sulphides, analytical uncertain-

(a)
ties and corrections for nonsulphide-hosted metals are gen-
30% or
Err erally unimportant. Corrections are likely to be trivial unless
Total combined error in
sulphide metal content

ti ve
+22.2%
Posi the sulphides are metal-poor, in which case the accuracy of
25%
rror
ati ve E rror the results is largely irrelevant, because all grade predictions
20% Neg itiveE
+12.2% Pos will be poor. However, overcorrection must be avoided,
%
±10 or preferably by the acquisition of “control data” from unmin-
15% -18.2% Err
a t ive eralized host rocks. Analytical errors can be a significant
Neg
10% ±2% problem if the samples are poor in sulphides.
-11.8%
5% The treatment of data from sulphide-poor silicate rocks
is inherently more complex. Correction for nonsulphide-
0%
0% 5% 10% 15% 20% hosted metals, and propagated analytical uncertainties, may
Analytical Precision for Metal become very significant. All these problems converge below
5% S, and may become critical below about 2.5% S. For
50% base metals, the calculation of sulphide metal contents from
(b)
samples containing < 1% S is not recommended, unless the
Total combined error in

host rocks can be shown to be devoid of metals, and non-

sulphide metal content

40%
30%
20%
10%
±4%
0%
0% 5% 10%
Analytical Precision for Metal
Figure 2. Illustration of compound errors in the calculation
of sulphide metal contents. (a) Positive (overestimate) and
negative (underestimate) errors due to analytical uncertain-
ties for metals ranging from ± 0% to ± 20%, with analytical
uncertainties of ± 2% and ± 10% for sulphur. (b) Average
combined uncertainties related to analytical precision for
metals and sulphur. Note that labels on the lines indicate
analytical precision for sulphur.
must be used in recalculation of any PGE data at low sul-
phide contents. There are also other reasons to be wary of
calculated sulphide PGE contents. The uncertainties implied
by Figure 2 must be borne in mind whenever data from dif-
ferent areas or mineralization types are compared and con-
trasted.
SUMMARY AND CONCLUSIONS
A comparison of techniques for calculation and correc-
tion of data to obtain sulphide metal contents shows no great
advantage for complex methods over simple ones, and the
latter are thus recommended in most situations. In mineral
exploration, where the prime objective is to assess the
±20% sulphide metal correction will generally be required below
±15% about 5% S. Correction is best approached through mass-
balance calculations, using input data from unmineralized
±10%
samples. It must always be remembered that undercorrec-
±6%
tion can cause false trends where metal contents are nega-
±2% tively correlated with the amount of sulphide, and overcor-
rection can generate false positive correlations. As the
amount of sulphide may also vary systematically (e.g.,
increasing toward the base of a magmatic body) such trends
15% 20% may be assigned unwarranted geological significance. The
distinction of true geological variation from this artificial
variation is perhaps the most difficult challenge in the inter-
pretation of data from sulphide-poor rocks.
Sulphur and metals are commonly analyzed from sepa-
rate aliquots, often at different laboratories, and sample het-
erogeneity may affect precision for all. Precision for sulphur
degrades below about 5% S, where nonsulphide metal cor-
rections become significant, and this adds to total uncertain-
ty at these levels. Base metals exhibit reasonable precision
(around ± 6%) in the range typical of most silicate-dominat-
ed rocks. Typical exploration-quality PGE data exhibit poor
precision unless sulphides are PGE-enriched, and the recal-
culation of such data from sulphide-poor samples is unwise.
For base metals, total combined errors in calculated sulphide
metal contents above 5% S are likely to be at least ± 10%,
and these will be worse at lower sulphur contents. These
propagated analytical errors must be considered whenever
sulphide metal contents are used to characterize different
“types” of mineralization, as subtle differences may be more
apparent than real. Sulphide PGE contents encounter a fur-
ther problem, because there is no guarantee that they are
hosted by common sulphides, or even by sulphides at all. In
rocks that have low S contents, such calculations may be
very misleading.

227
CURRENT RESEARCH, REPORT 03-1
To summarize, calculated sulphide metal contents have
unquestioned value in the evaluation and study of magmat-
ic sulphide deposits, and should be an important part of any
exploration program. Calculations are generally robust, but
results must be interpreted with knowledge of potential
problems, particularly in samples that are sulphide-poor.
These complications certainly do not preclude the use of the
method, and there are valid geological reasons for many
variations in sulphide metal contents described in the litera-
ture. Figures 1 and 2 provide guidelines for assessment of
correction problems and analytical error, and should prove
useful in assessing data of this type.
SPREADSHEETS FOR EASY
CALCULATION
The calculations involved in obtaining sulphide metal
contents are not complicated, but they are undeniably
tedious. Spreadsheet programs were developed for the cal-
culation and correction methods discussed in this article,
and are available as an open file (Kerr, 2003). The spread-
sheets are available in Lotus 1-2-3 and Microsoft Excel ver-
sions, and include documentation and help screens. A sepa-
rate spreadsheet is provided for the more complex method of
Naldrett et al. (2000). The spreadsheets are not designed for
the processing of large amounts of data, and present indi-
vidual records in column-format, rather than row-format,
because this makes it easier to compare uncorrected and cor-
rected data. However, they can easily be modified by users
to suit their specifications. The relevant open file can be
obtained on diskette or by e-mail from the Geological Sur-
vey, Geoscience Publications and Information Section
(www.geosurv.gov.nf.ca)
ACKNOWLEDGMENTS
This article was prompted largely by conversations with
exploration geologists in Labrador concerning such calcula-
tions. Chris Finch of the Department of Mines and Energy
laboratory is thanked for providing data on analytical preci-
sion and accuracy. Don James and Baxter Kean are thanked
for their comments on the initial version of this article.
REFERENCES
Barnes, S-J., Zientek, M.L. and Seversen, M.J.
1997: Ni, Cu, Au and platinum-group element contents
of sulphides associated with intraplate magmatism: a
synthesis. Canadian Journal of Earth Sciences, Volume
34, pages 337-351.
228
Finch, C.
1998: Inductively-coupled plasma emission spectrome-
try (ICP-ES) at the geochemical laboratory. In Current
Research. Newfoundland Department of Mines and
Energy, Geological Survey, Report 98-1, pages 179-
195.
Kerr, A.
1999: Mafic rocks of the Pants Lake Intrusion and relat-
ed Ni–Cu–Co mineralization in north-central Labrador.
In Current Research. Newfoundland Department of
Mines and Energy, Geological Survey, Report 99-1,
pages 215-253.
2002: A reconnaissance study of platinum-group-ele-
ment (PGE) contents from magmatic sulphide mineral-
ization in Labrador. In Current Research. Newfound-
land and Labrador Department of Mines and Energy,
Geological Survey, Report 99-1, pages 215-253.
2003: Spreadsheets for the calculation and correction of
sulphide metal contents. Newfoundland and Labrador
Department of Mines and Energy, Geological Survey,
Open File NFLD/2805.
In press: The calculation and use of sulphide metal con-
tents in the study of magmatic ore deposits: a method-
ological analysis. Exploration and Mining Geology.
Kerr, A. and Smith, J.L.
2000: Magmatic Ni–Cu sulphide mineralization in the
Harp Lake Intrusive Suite, central Labrador. In Current
Research. Newfoundland Department of Mines and
Energy, Geological Survey Branch, Report 2000-1,
pages 311-335.
Li, C., Maier, W.D. and de Waal, S.A.
2001: Magmatic Ni versus PGE deposits: Contrasting
genetic controls and exploration implications. South
African Journal of Geology, Volume 104, pages 309-
318.
Li, C. and Naldrett, A.J.
1999: Geology and petrology of the Voisey’s Bay Intru-
sion: reaction of olivine with silicate and sulfide liq-
uids. Lithos, Volume 47, pages 1-31.
Lightfoot, P.C., Keays, R.R. and Doherty, W.
2001a: Chemical evolution and origin of nickel sulfide
mineralization in the Sudbury Igneous Complex,
Ontario, Canada. Economic Geology, Volume 96, pages
1855-1873.

Lightfoot, P.C., Bell, C.C. and Muggridge, M.G.
2001b: Sulphide saturation history of the Voisey’s Bay
Intrusion, and formation of the nickel sulphide deposits.
Geological Association of Canada, Abstracts, Volume
26, pages 87.
A. KERR
Naldrett, A.J., Asif, M., Krstic, S. and Li, C.
2000: The composition of mineralization at the Voisey’s
Bay Ni–Cu sulfide deposit, with special reference to
platinum-group elements. Economic Geology, Volume
95, pages 845-867.
229